arch/x86/include/asm/kvm_host.h

   1 /*
   2  * Kernel-based Virtual Machine driver for Linux
   3  *
   4  * This header defines architecture specific interfaces, x86 version
   5  *
   6  * This work is licensed under the terms of the GNU GPL, version 2.  See
   7  * the COPYING file in the top-level directory.
   8  *
   9  */
  10
  11 #ifndef _ASM_X86_KVM_HOST_H
  12 #define _ASM_X86_KVM_HOST_H
  13
  14 #include <linux/types.h>
  15 #include <linux/mm.h>
  16 #include <linux/mmu_notifier.h>
  17 #include <linux/tracepoint.h>
  18 #include <linux/cpumask.h>
  19 #include <linux/irq_work.h>
  20
  21 #include <linux/kvm.h>
  22 #include <linux/kvm_para.h>
  23 #include <linux/kvm_types.h>
  24 #include <linux/perf_event.h>
  25 #include <linux/pvclock_gtod.h>
  26 #include <linux/clocksource.h>
  27
  28 #include <asm/pvclock-abi.h>
  29 #include <asm/desc.h>
  30 #include <asm/mtrr.h>
  31 #include <asm/msr-index.h>
  32 #include <asm/asm.h>
  33
  34 #define KVM_MAX_VCPUS 254
  35 #define KVM_SOFT_MAX_VCPUS 160
  36 #define KVM_USER_MEM_SLOTS 125
  37 /* memory slots that are not exposed to userspace */
  38 #define KVM_PRIVATE_MEM_SLOTS 3
  39 #define KVM_MEM_SLOTS_NUM (KVM_USER_MEM_SLOTS + KVM_PRIVATE_MEM_SLOTS)
  40
  41 #define KVM_MMIO_SIZE 16
  42
  43 #define KVM_PIO_PAGE_OFFSET 1
  44 #define KVM_COALESCED_MMIO_PAGE_OFFSET 2
  45
  46 #define CR0_RESERVED_BITS                                               \
  47         (~(unsigned long)(X86_CR0_PE | X86_CR0_MP | X86_CR0_EM | X86_CR0_TS \
  48                           | X86_CR0_ET | X86_CR0_NE | X86_CR0_WP | X86_CR0_AM \
  49                           | X86_CR0_NW | X86_CR0_CD | X86_CR0_PG))
  50
  51 #define CR3_PAE_RESERVED_BITS ((X86_CR3_PWT | X86_CR3_PCD) - 1)
  52 #define CR3_NONPAE_RESERVED_BITS ((PAGE_SIZE-1) & ~(X86_CR3_PWT | X86_CR3_PCD))
  53 #define CR3_PCID_ENABLED_RESERVED_BITS 0xFFFFFF0000000000ULL
  54 #define CR3_L_MODE_RESERVED_BITS (CR3_NONPAE_RESERVED_BITS |    \
  55                                   0xFFFFFF0000000000ULL)
  56 #define CR4_RESERVED_BITS                                               \
  57         (~(unsigned long)(X86_CR4_VME | X86_CR4_PVI | X86_CR4_TSD | X86_CR4_DE\
  58                           | X86_CR4_PSE | X86_CR4_PAE | X86_CR4_MCE     \
  59                           | X86_CR4_PGE | X86_CR4_PCE | X86_CR4_OSFXSR | X86_CR4_PCIDE \
  60                           | X86_CR4_OSXSAVE | X86_CR4_SMEP | X86_CR4_RDWRGSFS \
  61                           | X86_CR4_OSXMMEXCPT | X86_CR4_VMXE))
  62
  63 #define CR8_RESERVED_BITS (~(unsigned long)X86_CR8_TPR)
  64
  65
  66
  67 #define INVALID_PAGE (~(hpa_t)0)
  68 #define VALID_PAGE(x) ((x) != INVALID_PAGE)
  69
  70 #define UNMAPPED_GVA (~(gpa_t)0)
  71
  72 /* KVM Hugepage definitions for x86 */
  73 #define KVM_NR_PAGE_SIZES       3
  74 #define KVM_HPAGE_GFN_SHIFT(x)  (((x) - 1) * 9)
  75 #define KVM_HPAGE_SHIFT(x)      (PAGE_SHIFT + KVM_HPAGE_GFN_SHIFT(x))
  76 #define KVM_HPAGE_SIZE(x)       (1UL << KVM_HPAGE_SHIFT(x))
  77 #define KVM_HPAGE_MASK(x)       (~(KVM_HPAGE_SIZE(x) - 1))
  78 #define KVM_PAGES_PER_HPAGE(x)  (KVM_HPAGE_SIZE(x) / PAGE_SIZE)
  79
  80 #define SELECTOR_TI_MASK (1 << 2)
  81 #define SELECTOR_RPL_MASK 0x03
  82
  83 #define IOPL_SHIFT 12
  84
  85 #define KVM_PERMILLE_MMU_PAGES 20
  86 #define KVM_MIN_ALLOC_MMU_PAGES 64
  87 #define KVM_MMU_HASH_SHIFT 10
  88 #define KVM_NUM_MMU_PAGES (1 << KVM_MMU_HASH_SHIFT)
  89 #define KVM_MIN_FREE_MMU_PAGES 5
  90 #define KVM_REFILL_PAGES 25
  91 #define KVM_MAX_CPUID_ENTRIES 80
  92 #define KVM_NR_FIXED_MTRR_REGION 88
  93 #define KVM_NR_VAR_MTRR 8
  94
  95 #define ASYNC_PF_PER_VCPU 64
  96
  97 extern raw_spinlock_t kvm_lock;
  98 extern struct list_head vm_list;
  99
 100 struct kvm_vcpu;
 101 struct kvm;
 102 struct kvm_async_pf;
 103
 104 enum kvm_reg {
 105         VCPU_REGS_RAX = 0,
 106         VCPU_REGS_RCX = 1,
 107         VCPU_REGS_RDX = 2,
 108         VCPU_REGS_RBX = 3,
 109         VCPU_REGS_RSP = 4,
 110         VCPU_REGS_RBP = 5,
 111         VCPU_REGS_RSI = 6,
 112         VCPU_REGS_RDI = 7,
 113 #ifdef CONFIG_X86_64
 114         VCPU_REGS_R8 = 8,
 115         VCPU_REGS_R9 = 9,
 116         VCPU_REGS_R10 = 10,
 117         VCPU_REGS_R11 = 11,
 118         VCPU_REGS_R12 = 12,
 119         VCPU_REGS_R13 = 13,
 120         VCPU_REGS_R14 = 14,
 121         VCPU_REGS_R15 = 15,
 122 #endif
 123         VCPU_REGS_RIP,
 124         NR_VCPU_REGS
 125 };
 126
 127 enum kvm_reg_ex {
 128         VCPU_EXREG_PDPTR = NR_VCPU_REGS,
 129         VCPU_EXREG_CR3,
 130         VCPU_EXREG_RFLAGS,
 131         VCPU_EXREG_CPL,
 132         VCPU_EXREG_SEGMENTS,
 133 };
 134
 135 enum {
 136         VCPU_SREG_ES,
 137         VCPU_SREG_CS,
 138         VCPU_SREG_SS,
 139         VCPU_SREG_DS,
 140         VCPU_SREG_FS,
 141         VCPU_SREG_GS,
 142         VCPU_SREG_TR,
 143         VCPU_SREG_LDTR,
 144 };
 145
 146 #include <asm/kvm_emulate.h>
 147
 148 #define KVM_NR_MEM_OBJS 40
 149
 150 #define KVM_NR_DB_REGS  4
 151
 152 #define DR6_BD          (1 << 13)
 153 #define DR6_BS          (1 << 14)
 154 #define DR6_FIXED_1     0xffff0ff0
 155 #define DR6_VOLATILE    0x0000e00f
 156
 157 #define DR7_BP_EN_MASK  0x000000ff
 158 #define DR7_GE          (1 << 9)
 159 #define DR7_GD          (1 << 13)
 160 #define DR7_FIXED_1     0x00000400
 161 #define DR7_VOLATILE    0xffff23ff
 162
 163 /* apic attention bits */
 164 #define KVM_APIC_CHECK_VAPIC    0
 165 /*
 166  * The following bit is set with PV-EOI, unset on EOI.
 167  * We detect PV-EOI changes by guest by comparing
 168  * this bit with PV-EOI in guest memory.
 169  * See the implementation in apic_update_pv_eoi.
 170  */
 171 #define KVM_APIC_PV_EOI_PENDING 1
 172
 173 /*
 174  * We don't want allocation failures within the mmu code, so we preallocate
 175  * enough memory for a single page fault in a cache.
 176  */
 177 struct kvm_mmu_memory_cache {
 178         int nobjs;
 179         void *objects[KVM_NR_MEM_OBJS];
 180 };
 181
 182 /*
 183  * kvm_mmu_page_role, below, is defined as:
 184  *
 185  *   bits 0:3 - total guest paging levels (2-4, or zero for real mode)
 186  *   bits 4:7 - page table level for this shadow (1-4)
 187  *   bits 8:9 - page table quadrant for 2-level guests
 188  *   bit   16 - direct mapping of virtual to physical mapping at gfn
 189  *              used for real mode and two-dimensional paging
 190  *   bits 17:19 - common access permissions for all ptes in this shadow page
 191  */
 192 union kvm_mmu_page_role {
 193         unsigned word;
 194         struct {
 195                 unsigned level:4;
 196                 unsigned cr4_pae:1;
 197                 unsigned quadrant:2;
 198                 unsigned pad_for_nice_hex_output:6;
 199                 unsigned direct:1;
 200                 unsigned access:3;
 201                 unsigned invalid:1;
 202                 unsigned nxe:1;
 203                 unsigned cr0_wp:1;
 204                 unsigned smep_andnot_wp:1;
 205         };
 206 };
 207
 208 struct kvm_mmu_page {
 209         struct list_head link;
 210         struct hlist_node hash_link;
 211
 212         /*
 213          * The following two entries are used to key the shadow page in the
 214          * hash table.
 215          */
 216         gfn_t gfn;
 217         union kvm_mmu_page_role role;
 218
 219         u64 *spt;
 220         /* hold the gfn of each spte inside spt */
 221         gfn_t *gfns;
 222         bool unsync;
 223         int root_count;          /* Currently serving as active root */
 224         unsigned int unsync_children;
 225         unsigned long parent_ptes;      /* Reverse mapping for parent_pte */
 226         DECLARE_BITMAP(unsync_child_bitmap, 512);
 227
 228 #ifdef CONFIG_X86_32
 229         int clear_spte_count;
 230 #endif
 231
 232         int write_flooding_count;
 233 };
 234
 235 struct kvm_pio_request {
 236         unsigned long count;
 237         int in;
 238         int port;
 239         int size;
 240 };
 241
 242 /*
 243  * x86 supports 3 paging modes (4-level 64-bit, 3-level 64-bit, and 2-level
 244  * 32-bit).  The kvm_mmu structure abstracts the details of the current mmu
 245  * mode.
 246  */
 247 struct kvm_mmu {
 248         void (*new_cr3)(struct kvm_vcpu *vcpu);
 249         void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long root);
 250         unsigned long (*get_cr3)(struct kvm_vcpu *vcpu);
 251         u64 (*get_pdptr)(struct kvm_vcpu *vcpu, int index);
 252         int (*page_fault)(struct kvm_vcpu *vcpu, gva_t gva, u32 err,
 253                           bool prefault);
 254         void (*inject_page_fault)(struct kvm_vcpu *vcpu,
 255                                   struct x86_exception *fault);
 256         void (*free)(struct kvm_vcpu *vcpu);
 257         gpa_t (*gva_to_gpa)(struct kvm_vcpu *vcpu, gva_t gva, u32 access,
 258                             struct x86_exception *exception);
 259         gpa_t (*translate_gpa)(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
 260         int (*sync_page)(struct kvm_vcpu *vcpu,
 261                          struct kvm_mmu_page *sp);
 262         void (*invlpg)(struct kvm_vcpu *vcpu, gva_t gva);
 263         void (*update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
 264                            u64 *spte, const void *pte);
 265         hpa_t root_hpa;
 266         int root_level;
 267         int shadow_root_level;
 268         union kvm_mmu_page_role base_role;
 269         bool direct_map;
 270
 271         /*
 272          * Bitmap; bit set = permission fault
 273          * Byte index: page fault error code [4:1]
 274          * Bit index: pte permissions in ACC_* format
 275          */
 276         u8 permissions[16];
 277
 278         u64 *pae_root;
 279         u64 *lm_root;
 280         u64 rsvd_bits_mask[2][4];
 281
 282         /*
 283          * Bitmap: bit set = last pte in walk
 284          * index[0:1]: level (zero-based)
 285          * index[2]: pte.ps
 286          */
 287         u8 last_pte_bitmap;
 288
 289         bool nx;
 290
 291         u64 pdptrs[4]; /* pae */
 292 };
 293
 294 enum pmc_type {
 295         KVM_PMC_GP = 0,
 296         KVM_PMC_FIXED,
 297 };
 298
 299 struct kvm_pmc {
 300         enum pmc_type type;
 301         u8 idx;
 302         u64 counter;
 303         u64 eventsel;
 304         struct perf_event *perf_event;
 305         struct kvm_vcpu *vcpu;
 306 };
 307
 308 struct kvm_pmu {
 309         unsigned nr_arch_gp_counters;
 310         unsigned nr_arch_fixed_counters;
 311         unsigned available_event_types;
 312         u64 fixed_ctr_ctrl;
 313         u64 global_ctrl;
 314         u64 global_status;
 315         u64 global_ovf_ctrl;
 316         u64 counter_bitmask[2];
 317         u64 global_ctrl_mask;
 318         u8 version;
 319         struct kvm_pmc gp_counters[INTEL_PMC_MAX_GENERIC];
 320         struct kvm_pmc fixed_counters[INTEL_PMC_MAX_FIXED];
 321         struct irq_work irq_work;
 322         u64 reprogram_pmi;
 323 };
 324
 325 struct kvm_vcpu_arch {
 326         /*
 327          * rip and regs accesses must go through
 328          * kvm_{register,rip}_{read,write} functions.
 329          */
 330         unsigned long regs[NR_VCPU_REGS];
 331         u32 regs_avail;
 332         u32 regs_dirty;
 333
 334         unsigned long cr0;
 335         unsigned long cr0_guest_owned_bits;
 336         unsigned long cr2;
 337         unsigned long cr3;
 338         unsigned long cr4;
 339         unsigned long cr4_guest_owned_bits;
 340         unsigned long cr8;
 341         u32 hflags;
 342         u64 efer;
 343         u64 apic_base;
 344         struct kvm_lapic *apic;    /* kernel irqchip context */
 345         unsigned long apic_attention;
 346         int32_t apic_arb_prio;
 347         int mp_state;
 348         int sipi_vector;
 349         u64 ia32_misc_enable_msr;
 350         bool tpr_access_reporting;
 351
 352         /*
 353          * Paging state of the vcpu
 354          *
 355          * If the vcpu runs in guest mode with two level paging this still saves
 356          * the paging mode of the l1 guest. This context is always used to
 357          * handle faults.
 358          */
 359         struct kvm_mmu mmu;
 360
 361         /*
 362          * Paging state of an L2 guest (used for nested npt)
 363          *
 364          * This context will save all necessary information to walk page tables
 365          * of the an L2 guest. This context is only initialized for page table
 366          * walking and not for faulting since we never handle l2 page faults on
 367          * the host.
 368          */
 369         struct kvm_mmu nested_mmu;
 370
 371         /*
 372          * Pointer to the mmu context currently used for
 373          * gva_to_gpa translations.
 374          */
 375         struct kvm_mmu *walk_mmu;
 376
 377         struct kvm_mmu_memory_cache mmu_pte_list_desc_cache;
 378         struct kvm_mmu_memory_cache mmu_page_cache;
 379         struct kvm_mmu_memory_cache mmu_page_header_cache;
 380
 381         struct fpu guest_fpu;
 382         u64 xcr0;
 383
 384         struct kvm_pio_request pio;
 385         void *pio_data;
 386
 387         u8 event_exit_inst_len;
 388
 389         struct kvm_queued_exception {
 390                 bool pending;
 391                 bool has_error_code;
 392                 bool reinject;
 393                 u8 nr;
 394                 u32 error_code;
 395         } exception;
 396
 397         struct kvm_queued_interrupt {
 398                 bool pending;
 399                 bool soft;
 400                 u8 nr;
 401         } interrupt;
 402
 403         int halt_request; /* real mode on Intel only */
 404
 405         int cpuid_nent;
 406         struct kvm_cpuid_entry2 cpuid_entries[KVM_MAX_CPUID_ENTRIES];
 407         /* emulate context */
 408
 409         struct x86_emulate_ctxt emulate_ctxt;
 410         bool emulate_regs_need_sync_to_vcpu;
 411         bool emulate_regs_need_sync_from_vcpu;
 412         int (*complete_userspace_io)(struct kvm_vcpu *vcpu);
 413
 414         gpa_t time;
 415         struct pvclock_vcpu_time_info hv_clock;
 416         unsigned int hw_tsc_khz;
 417         struct gfn_to_hva_cache pv_time;
 418         bool pv_time_enabled;
 419         /* set guest stopped flag in pvclock flags field */
 420         bool pvclock_set_guest_stopped_request;
 421
 422         struct {
 423                 u64 msr_val;
 424                 u64 last_steal;
 425                 u64 accum_steal;
 426                 struct gfn_to_hva_cache stime;
 427                 struct kvm_steal_time steal;
 428         } st;
 429
 430         u64 last_guest_tsc;
 431         u64 last_kernel_ns;
 432         u64 last_host_tsc;
 433         u64 tsc_offset_adjustment;
 434         u64 this_tsc_nsec;
 435         u64 this_tsc_write;
 436         u8  this_tsc_generation;
 437         bool tsc_catchup;
 438         bool tsc_always_catchup;
 439         s8 virtual_tsc_shift;
 440         u32 virtual_tsc_mult;
 441         u32 virtual_tsc_khz;
 442         s64 ia32_tsc_adjust_msr;
 443
 444         atomic_t nmi_queued;  /* unprocessed asynchronous NMIs */
 445         unsigned nmi_pending; /* NMI queued after currently running handler */
 446         bool nmi_injected;    /* Trying to inject an NMI this entry */
 447
 448         struct mtrr_state_type mtrr_state;
 449         u32 pat;
 450
 451         int switch_db_regs;
 452         unsigned long db[KVM_NR_DB_REGS];
 453         unsigned long dr6;
 454         unsigned long dr7;
 455         unsigned long eff_db[KVM_NR_DB_REGS];
 456         unsigned long guest_debug_dr7;
 457
 458         u64 mcg_cap;
 459         u64 mcg_status;
 460         u64 mcg_ctl;
 461         u64 *mce_banks;
 462
 463         /* Cache MMIO info */
 464         u64 mmio_gva;
 465         unsigned access;
 466         gfn_t mmio_gfn;
 467
 468         struct kvm_pmu pmu;
 469
 470         /* used for guest single stepping over the given code position */
 471         unsigned long singlestep_rip;
 472
 473         /* fields used by HYPER-V emulation */
 474         u64 hv_vapic;
 475
 476         cpumask_var_t wbinvd_dirty_mask;
 477
 478         unsigned long last_retry_eip;
 479         unsigned long last_retry_addr;
 480
 481         struct {
 482                 bool halted;
 483                 gfn_t gfns[roundup_pow_of_two(ASYNC_PF_PER_VCPU)];
 484                 struct gfn_to_hva_cache data;
 485                 u64 msr_val;
 486                 u32 id;
 487                 bool send_user_only;
 488         } apf;
 489
 490         /* OSVW MSRs (AMD only) */
 491         struct {
 492                 u64 length;
 493                 u64 status;
 494         } osvw;
 495
 496         struct {
 497                 u64 msr_val;
 498                 struct gfn_to_hva_cache data;
 499         } pv_eoi;
 500
 501         /*
 502          * Indicate whether the access faults on its page table in guest
 503          * which is set when fix page fault and used to detect unhandeable
 504          * instruction.
 505          */
 506         bool write_fault_to_shadow_pgtable;
 507 };
 508
 509 struct kvm_lpage_info {
 510         int write_count;
 511 };
 512
 513 struct kvm_arch_memory_slot {
 514         unsigned long *rmap[KVM_NR_PAGE_SIZES];
 515         struct kvm_lpage_info *lpage_info[KVM_NR_PAGE_SIZES - 1];
 516 };
 517
 518 struct kvm_apic_map {
 519         struct rcu_head rcu;
 520         u8 ldr_bits;
 521         /* fields bellow are used to decode ldr values in different modes */
 522         u32 cid_shift, cid_mask, lid_mask;
 523         struct kvm_lapic *phys_map[256];
 524         /* first index is cluster id second is cpu id in a cluster */
 525         struct kvm_lapic *logical_map[16][16];
 526 };
 527
 528 struct kvm_arch {
 529         unsigned int n_used_mmu_pages;
 530         unsigned int n_requested_mmu_pages;
 531         unsigned int n_max_mmu_pages;
 532         unsigned int indirect_shadow_pages;
 533         struct hlist_head mmu_page_hash[KVM_NUM_MMU_PAGES];
 534         /*
 535          * Hash table of struct kvm_mmu_page.
 536          */
 537         struct list_head active_mmu_pages;
 538         struct list_head assigned_dev_head;
 539         struct iommu_domain *iommu_domain;
 540         int iommu_flags;
 541         struct kvm_pic *vpic;
 542         struct kvm_ioapic *vioapic;
 543         struct kvm_pit *vpit;
 544         int vapics_in_nmi_mode;
 545         struct mutex apic_map_lock;
 546         struct kvm_apic_map *apic_map;
 547
 548         unsigned int tss_addr;
 549         struct page *apic_access_page;
 550
 551         gpa_t wall_clock;
 552
 553         struct page *ept_identity_pagetable;
 554         bool ept_identity_pagetable_done;
 555         gpa_t ept_identity_map_addr;
 556
 557         unsigned long irq_sources_bitmap;
 558         s64 kvmclock_offset;
 559         raw_spinlock_t tsc_write_lock;
 560         u64 last_tsc_nsec;
 561         u64 last_tsc_write;
 562         u32 last_tsc_khz;
 563         u64 cur_tsc_nsec;
 564         u64 cur_tsc_write;
 565         u64 cur_tsc_offset;
 566         u8  cur_tsc_generation;
 567         int nr_vcpus_matched_tsc;
 568
 569         spinlock_t pvclock_gtod_sync_lock;
 570         bool use_master_clock;
 571         u64 master_kernel_ns;
 572         cycle_t master_cycle_now;
 573
 574         struct kvm_xen_hvm_config xen_hvm_config;
 575
 576         /* fields used by HYPER-V emulation */
 577         u64 hv_guest_os_id;
 578         u64 hv_hypercall;
 579
 580         #ifdef CONFIG_KVM_MMU_AUDIT
 581         int audit_point;
 582         #endif
 583 };
 584
 585 struct kvm_vm_stat {
 586         u32 mmu_shadow_zapped;
 587         u32 mmu_pte_write;
 588         u32 mmu_pte_updated;
 589         u32 mmu_pde_zapped;
 590         u32 mmu_flooded;
 591         u32 mmu_recycled;
 592         u32 mmu_cache_miss;
 593         u32 mmu_unsync;
 594         u32 remote_tlb_flush;
 595         u32 lpages;
 596 };
 597
 598 struct kvm_vcpu_stat {
 599         u32 pf_fixed;
 600         u32 pf_guest;
 601         u32 tlb_flush;
 602         u32 invlpg;
 603
 604         u32 exits;
 605         u32 io_exits;
 606         u32 mmio_exits;
 607         u32 signal_exits;
 608         u32 irq_window_exits;
 609         u32 nmi_window_exits;
 610         u32 halt_exits;
 611         u32 halt_wakeup;
 612         u32 request_irq_exits;
 613         u32 irq_exits;
 614         u32 host_state_reload;
 615         u32 efer_reload;
 616         u32 fpu_reload;
 617         u32 insn_emulation;
 618         u32 insn_emulation_fail;
 619         u32 hypercalls;
 620         u32 irq_injections;
 621         u32 nmi_injections;
 622 };
 623
 624 struct x86_instruction_info;
 625
 626 struct msr_data {
 627         bool host_initiated;
 628         u32 index;
 629         u64 data;
 630 };
 631
 632 struct kvm_x86_ops {
 633         int (*cpu_has_kvm_support)(void);          /* __init */
 634         int (*disabled_by_bios)(void);             /* __init */
 635         int (*hardware_enable)(void *dummy);
 636         void (*hardware_disable)(void *dummy);
 637         void (*check_processor_compatibility)(void *rtn);
 638         int (*hardware_setup)(void);               /* __init */
 639         void (*hardware_unsetup)(void);            /* __exit */
 640         bool (*cpu_has_accelerated_tpr)(void);
 641         void (*cpuid_update)(struct kvm_vcpu *vcpu);
 642
 643         /* Create, but do not attach this VCPU */
 644         struct kvm_vcpu *(*vcpu_create)(struct kvm *kvm, unsigned id);
 645         void (*vcpu_free)(struct kvm_vcpu *vcpu);
 646         int (*vcpu_reset)(struct kvm_vcpu *vcpu);
 647
 648         void (*prepare_guest_switch)(struct kvm_vcpu *vcpu);
 649         void (*vcpu_load)(struct kvm_vcpu *vcpu, int cpu);
 650         void (*vcpu_put)(struct kvm_vcpu *vcpu);
 651
 652         void (*update_db_bp_intercept)(struct kvm_vcpu *vcpu);
 653         int (*get_msr)(struct kvm_vcpu *vcpu, u32 msr_index, u64 *pdata);
 654         int (*set_msr)(struct kvm_vcpu *vcpu, struct msr_data *msr);
 655         u64 (*get_segment_base)(struct kvm_vcpu *vcpu, int seg);
 656         void (*get_segment)(struct kvm_vcpu *vcpu,
 657                             struct kvm_segment *var, int seg);
 658         int (*get_cpl)(struct kvm_vcpu *vcpu);
 659         void (*set_segment)(struct kvm_vcpu *vcpu,
 660                             struct kvm_segment *var, int seg);
 661         void (*get_cs_db_l_bits)(struct kvm_vcpu *vcpu, int *db, int *l);
 662         void (*decache_cr0_guest_bits)(struct kvm_vcpu *vcpu);
 663         void (*decache_cr3)(struct kvm_vcpu *vcpu);
 664         void (*decache_cr4_guest_bits)(struct kvm_vcpu *vcpu);
 665         void (*set_cr0)(struct kvm_vcpu *vcpu, unsigned long cr0);
 666         void (*set_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
 667         int (*set_cr4)(struct kvm_vcpu *vcpu, unsigned long cr4);
 668         void (*set_efer)(struct kvm_vcpu *vcpu, u64 efer);
 669         void (*get_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 670         void (*set_idt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 671         void (*get_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 672         void (*set_gdt)(struct kvm_vcpu *vcpu, struct desc_ptr *dt);
 673         void (*set_dr7)(struct kvm_vcpu *vcpu, unsigned long value);
 674         void (*cache_reg)(struct kvm_vcpu *vcpu, enum kvm_reg reg);
 675         unsigned long (*get_rflags)(struct kvm_vcpu *vcpu);
 676         void (*set_rflags)(struct kvm_vcpu *vcpu, unsigned long rflags);
 677         void (*fpu_activate)(struct kvm_vcpu *vcpu);
 678         void (*fpu_deactivate)(struct kvm_vcpu *vcpu);
 679
 680         void (*tlb_flush)(struct kvm_vcpu *vcpu);
 681
 682         void (*run)(struct kvm_vcpu *vcpu);
 683         int (*handle_exit)(struct kvm_vcpu *vcpu);
 684         void (*skip_emulated_instruction)(struct kvm_vcpu *vcpu);
 685         void (*set_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
 686         u32 (*get_interrupt_shadow)(struct kvm_vcpu *vcpu, int mask);
 687         void (*patch_hypercall)(struct kvm_vcpu *vcpu,
 688                                 unsigned char *hypercall_addr);
 689         void (*set_irq)(struct kvm_vcpu *vcpu);
 690         void (*set_nmi)(struct kvm_vcpu *vcpu);
 691         void (*queue_exception)(struct kvm_vcpu *vcpu, unsigned nr,
 692                                 bool has_error_code, u32 error_code,
 693                                 bool reinject);
 694         void (*cancel_injection)(struct kvm_vcpu *vcpu);
 695         int (*interrupt_allowed)(struct kvm_vcpu *vcpu);
 696         int (*nmi_allowed)(struct kvm_vcpu *vcpu);
 697         bool (*get_nmi_mask)(struct kvm_vcpu *vcpu);
 698         void (*set_nmi_mask)(struct kvm_vcpu *vcpu, bool masked);
 699         void (*enable_nmi_window)(struct kvm_vcpu *vcpu);
 700         void (*enable_irq_window)(struct kvm_vcpu *vcpu);
 701         void (*update_cr8_intercept)(struct kvm_vcpu *vcpu, int tpr, int irr);
 702         int (*vm_has_apicv)(struct kvm *kvm);
 703         void (*hwapic_irr_update)(struct kvm_vcpu *vcpu, int max_irr);
 704         void (*hwapic_isr_update)(struct kvm *kvm, int isr);
 705         void (*load_eoi_exitmap)(struct kvm_vcpu *vcpu, u64 *eoi_exit_bitmap);
 706         void (*set_virtual_x2apic_mode)(struct kvm_vcpu *vcpu, bool set);
 707         int (*set_tss_addr)(struct kvm *kvm, unsigned int addr);
 708         int (*get_tdp_level)(void);
 709         u64 (*get_mt_mask)(struct kvm_vcpu *vcpu, gfn_t gfn, bool is_mmio);
 710         int (*get_lpage_level)(void);
 711         bool (*rdtscp_supported)(void);
 712         bool (*invpcid_supported)(void);
 713         void (*adjust_tsc_offset)(struct kvm_vcpu *vcpu, s64 adjustment, bool host);
 714
 715         void (*set_tdp_cr3)(struct kvm_vcpu *vcpu, unsigned long cr3);
 716
 717         void (*set_supported_cpuid)(u32 func, struct kvm_cpuid_entry2 *entry);
 718
 719         bool (*has_wbinvd_exit)(void);
 720
 721         void (*set_tsc_khz)(struct kvm_vcpu *vcpu, u32 user_tsc_khz, bool scale);
 722         u64 (*read_tsc_offset)(struct kvm_vcpu *vcpu);
 723         void (*write_tsc_offset)(struct kvm_vcpu *vcpu, u64 offset);
 724
 725         u64 (*compute_tsc_offset)(struct kvm_vcpu *vcpu, u64 target_tsc);
 726         u64 (*read_l1_tsc)(struct kvm_vcpu *vcpu, u64 host_tsc);
 727
 728         void (*get_exit_info)(struct kvm_vcpu *vcpu, u64 *info1, u64 *info2);
 729
 730         int (*check_intercept)(struct kvm_vcpu *vcpu,
 731                                struct x86_instruction_info *info,
 732                                enum x86_intercept_stage stage);
 733 };
 734
 735 struct kvm_arch_async_pf {
 736         u32 token;
 737         gfn_t gfn;
 738         unsigned long cr3;
 739         bool direct_map;
 740 };
 741
 742 extern struct kvm_x86_ops *kvm_x86_ops;
 743
 744 static inline void adjust_tsc_offset_guest(struct kvm_vcpu *vcpu,
 745                                            s64 adjustment)
 746 {
 747         kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, false);
 748 }
 749
 750 static inline void adjust_tsc_offset_host(struct kvm_vcpu *vcpu, s64 adjustment)
 751 {
 752         kvm_x86_ops->adjust_tsc_offset(vcpu, adjustment, true);
 753 }
 754
 755 int kvm_mmu_module_init(void);
 756 void kvm_mmu_module_exit(void);
 757
 758 void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
 759 int kvm_mmu_create(struct kvm_vcpu *vcpu);
 760 int kvm_mmu_setup(struct kvm_vcpu *vcpu);
 761 void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
 762                 u64 dirty_mask, u64 nx_mask, u64 x_mask);
 763
 764 int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
 765 void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
 766 void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
 767                                      struct kvm_memory_slot *slot,
 768                                      gfn_t gfn_offset, unsigned long mask);
 769 void kvm_mmu_zap_all(struct kvm *kvm);
 770 unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm);
 771 void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int kvm_nr_mmu_pages);
 772
 773 int load_pdptrs(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu, unsigned long cr3);
 774
 775 int emulator_write_phys(struct kvm_vcpu *vcpu, gpa_t gpa,
 776                           const void *val, int bytes);
 777 u8 kvm_get_guest_memory_type(struct kvm_vcpu *vcpu, gfn_t gfn);
 778
 779 extern bool tdp_enabled;
 780
 781 u64 vcpu_tsc_khz(struct kvm_vcpu *vcpu);
 782
 783 /* control of guest tsc rate supported? */
 784 extern bool kvm_has_tsc_control;
 785 /* minimum supported tsc_khz for guests */
 786 extern u32  kvm_min_guest_tsc_khz;
 787 /* maximum supported tsc_khz for guests */
 788 extern u32  kvm_max_guest_tsc_khz;
 789
 790 enum emulation_result {
 791         EMULATE_DONE,       /* no further processing */
 792         EMULATE_DO_MMIO,      /* kvm_run filled with mmio request */
 793         EMULATE_FAIL,         /* can't emulate this instruction */
 794 };
 795
 796 #define EMULTYPE_NO_DECODE          (1 << 0)
 797 #define EMULTYPE_TRAP_UD            (1 << 1)
 798 #define EMULTYPE_SKIP               (1 << 2)
 799 #define EMULTYPE_RETRY              (1 << 3)
 800 int x86_emulate_instruction(struct kvm_vcpu *vcpu, unsigned long cr2,
 801                             int emulation_type, void *insn, int insn_len);
 802
 803 static inline int emulate_instruction(struct kvm_vcpu *vcpu,
 804                         int emulation_type)
 805 {
 806         return x86_emulate_instruction(vcpu, 0, emulation_type, NULL, 0);
 807 }
 808
 809 void kvm_enable_efer_bits(u64);
 810 int kvm_get_msr(struct kvm_vcpu *vcpu, u32 msr_index, u64 *data);
 811 int kvm_set_msr(struct kvm_vcpu *vcpu, struct msr_data *msr);
 812
 813 struct x86_emulate_ctxt;
 814
 815 int kvm_fast_pio_out(struct kvm_vcpu *vcpu, int size, unsigned short port);
 816 void kvm_emulate_cpuid(struct kvm_vcpu *vcpu);
 817 int kvm_emulate_halt(struct kvm_vcpu *vcpu);
 818 int kvm_emulate_wbinvd(struct kvm_vcpu *vcpu);
 819
 820 void kvm_get_segment(struct kvm_vcpu *vcpu, struct kvm_segment *var, int seg);
 821 int kvm_load_segment_descriptor(struct kvm_vcpu *vcpu, u16 selector, int seg);
 822
 823 int kvm_task_switch(struct kvm_vcpu *vcpu, u16 tss_selector, int idt_index,
 824                     int reason, bool has_error_code, u32 error_code);
 825
 826 int kvm_set_cr0(struct kvm_vcpu *vcpu, unsigned long cr0);
 827 int kvm_set_cr3(struct kvm_vcpu *vcpu, unsigned long cr3);
 828 int kvm_set_cr4(struct kvm_vcpu *vcpu, unsigned long cr4);
 829 int kvm_set_cr8(struct kvm_vcpu *vcpu, unsigned long cr8);
 830 int kvm_set_dr(struct kvm_vcpu *vcpu, int dr, unsigned long val);
 831 int kvm_get_dr(struct kvm_vcpu *vcpu, int dr, unsigned long *val);
 832 unsigned long kvm_get_cr8(struct kvm_vcpu *vcpu);
 833 void kvm_lmsw(struct kvm_vcpu *vcpu, unsigned long msw);
 834 void kvm_get_cs_db_l_bits(struct kvm_vcpu *vcpu, int *db, int *l);
 835 int kvm_set_xcr(struct kvm_vcpu *vcpu, u32 index, u64 xcr);
 836
 837 int kvm_get_msr_common(struct kvm_vcpu *vcpu, u32 msr, u64 *pdata);
 838 int kvm_set_msr_common(struct kvm_vcpu *vcpu, struct msr_data *msr);
 839
 840 unsigned long kvm_get_rflags(struct kvm_vcpu *vcpu);
 841 void kvm_set_rflags(struct kvm_vcpu *vcpu, unsigned long rflags);
 842 bool kvm_rdpmc(struct kvm_vcpu *vcpu);
 843
 844 void kvm_queue_exception(struct kvm_vcpu *vcpu, unsigned nr);
 845 void kvm_queue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
 846 void kvm_requeue_exception(struct kvm_vcpu *vcpu, unsigned nr);
 847 void kvm_requeue_exception_e(struct kvm_vcpu *vcpu, unsigned nr, u32 error_code);
 848 void kvm_inject_page_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
 849 int kvm_read_guest_page_mmu(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
 850                             gfn_t gfn, void *data, int offset, int len,
 851                             u32 access);
 852 void kvm_propagate_fault(struct kvm_vcpu *vcpu, struct x86_exception *fault);
 853 bool kvm_require_cpl(struct kvm_vcpu *vcpu, int required_cpl);
 854
 855 static inline int __kvm_irq_line_state(unsigned long *irq_state,
 856                                        int irq_source_id, int level)
 857 {
 858         /* Logical OR for level trig interrupt */
 859         if (level)
 860                 __set_bit(irq_source_id, irq_state);
 861         else
 862                 __clear_bit(irq_source_id, irq_state);
 863
 864         return !!(*irq_state);
 865 }
 866
 867 int kvm_pic_set_irq(struct kvm_pic *pic, int irq, int irq_source_id, int level);
 868 void kvm_pic_clear_all(struct kvm_pic *pic, int irq_source_id);
 869
 870 void kvm_inject_nmi(struct kvm_vcpu *vcpu);
 871
 872 int fx_init(struct kvm_vcpu *vcpu);
 873
 874 void kvm_mmu_flush_tlb(struct kvm_vcpu *vcpu);
 875 void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
 876                        const u8 *new, int bytes);
 877 int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn);
 878 int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva);
 879 void __kvm_mmu_free_some_pages(struct kvm_vcpu *vcpu);
 880 int kvm_mmu_load(struct kvm_vcpu *vcpu);
 881 void kvm_mmu_unload(struct kvm_vcpu *vcpu);
 882 void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu);
 883 gpa_t translate_nested_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access);
 884 gpa_t kvm_mmu_gva_to_gpa_read(struct kvm_vcpu *vcpu, gva_t gva,
 885                               struct x86_exception *exception);
 886 gpa_t kvm_mmu_gva_to_gpa_fetch(struct kvm_vcpu *vcpu, gva_t gva,
 887                                struct x86_exception *exception);
 888 gpa_t kvm_mmu_gva_to_gpa_write(struct kvm_vcpu *vcpu, gva_t gva,
 889                                struct x86_exception *exception);
 890 gpa_t kvm_mmu_gva_to_gpa_system(struct kvm_vcpu *vcpu, gva_t gva,
 891                                 struct x86_exception *exception);
 892
 893 int kvm_emulate_hypercall(struct kvm_vcpu *vcpu);
 894
 895 int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t gva, u32 error_code,
 896                        void *insn, int insn_len);
 897 void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva);
 898
 899 void kvm_enable_tdp(void);
 900 void kvm_disable_tdp(void);
 901
 902 int complete_pio(struct kvm_vcpu *vcpu);
 903 bool kvm_check_iopl(struct kvm_vcpu *vcpu);
 904
 905 static inline gpa_t translate_gpa(struct kvm_vcpu *vcpu, gpa_t gpa, u32 access)
 906 {
 907         return gpa;
 908 }
 909
 910 static inline struct kvm_mmu_page *page_header(hpa_t shadow_page)
 911 {
 912         struct page *page = pfn_to_page(shadow_page >> PAGE_SHIFT);
 913
 914         return (struct kvm_mmu_page *)page_private(page);
 915 }
 916
 917 static inline u16 kvm_read_ldt(void)
 918 {
 919         u16 ldt;
 920         asm("sldt %0" : "=g"(ldt));
 921         return ldt;
 922 }
 923
 924 static inline void kvm_load_ldt(u16 sel)
 925 {
 926         asm("lldt %0" : : "rm"(sel));
 927 }
 928
 929 #ifdef CONFIG_X86_64
 930 static inline unsigned long read_msr(unsigned long msr)
 931 {
 932         u64 value;
 933
 934         rdmsrl(msr, value);
 935         return value;
 936 }
 937 #endif
 938
 939 static inline u32 get_rdx_init_val(void)
 940 {
 941         return 0x600; /* P6 family */
 942 }
 943
 944 static inline void kvm_inject_gp(struct kvm_vcpu *vcpu, u32 error_code)
 945 {
 946         kvm_queue_exception_e(vcpu, GP_VECTOR, error_code);
 947 }
 948
 949 #define TSS_IOPB_BASE_OFFSET 0x66
 950 #define TSS_BASE_SIZE 0x68
 951 #define TSS_IOPB_SIZE (65536 / 8)
 952 #define TSS_REDIRECTION_SIZE (256 / 8)
 953 #define RMODE_TSS_SIZE                                                  \
 954         (TSS_BASE_SIZE + TSS_REDIRECTION_SIZE + TSS_IOPB_SIZE + 1)
 955
 956 enum {
 957         TASK_SWITCH_CALL = 0,
 958         TASK_SWITCH_IRET = 1,
 959         TASK_SWITCH_JMP = 2,
 960         TASK_SWITCH_GATE = 3,
 961 };
 962
 963 #define HF_GIF_MASK             (1 << 0)
 964 #define HF_HIF_MASK             (1 << 1)
 965 #define HF_VINTR_MASK           (1 << 2)
 966 #define HF_NMI_MASK             (1 << 3)
 967 #define HF_IRET_MASK            (1 << 4)
 968 #define HF_GUEST_MASK           (1 << 5) /* VCPU is in guest-mode */
 969
 970 /*
 971  * Hardware virtualization extension instructions may fault if a
 972  * reboot turns off virtualization while processes are running.
 973  * Trap the fault and ignore the instruction if that happens.
 974  */
 975 asmlinkage void kvm_spurious_fault(void);
 976 extern bool kvm_rebooting;
 977
 978 #define ____kvm_handle_fault_on_reboot(insn, cleanup_insn)      \
 979         "666: " insn "\n\t" \
 980         "668: \n\t"                           \
 981         ".pushsection .fixup, \"ax\" \n" \
 982         "667: \n\t" \
 983         cleanup_insn "\n\t"                   \
 984         "cmpb $0, kvm_rebooting \n\t"         \
 985         "jne 668b \n\t"                       \
 986         __ASM_SIZE(push) " $666b \n\t"        \
 987         "call kvm_spurious_fault \n\t"        \
 988         ".popsection \n\t" \
 989         _ASM_EXTABLE(666b, 667b)
 990
 991 #define __kvm_handle_fault_on_reboot(insn)              \
 992         ____kvm_handle_fault_on_reboot(insn, "")
 993
 994 #define KVM_ARCH_WANT_MMU_NOTIFIER
 995 int kvm_unmap_hva(struct kvm *kvm, unsigned long hva);
 996 int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end);
 997 int kvm_age_hva(struct kvm *kvm, unsigned long hva);
 998 int kvm_test_age_hva(struct kvm *kvm, unsigned long hva);
 999 void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte);
1000 int cpuid_maxphyaddr(struct kvm_vcpu *vcpu);
1001 int kvm_cpu_has_injectable_intr(struct kvm_vcpu *v);
1002 int kvm_cpu_has_interrupt(struct kvm_vcpu *vcpu);
1003 int kvm_arch_interrupt_allowed(struct kvm_vcpu *vcpu);
1004 int kvm_cpu_get_interrupt(struct kvm_vcpu *v);
1005
1006 void kvm_define_shared_msr(unsigned index, u32 msr);
1007 void kvm_set_shared_msr(unsigned index, u64 val, u64 mask);
1008
1009 bool kvm_is_linear_rip(struct kvm_vcpu *vcpu, unsigned long linear_rip);
1010
1011 void kvm_arch_async_page_not_present(struct kvm_vcpu *vcpu,
1012                                      struct kvm_async_pf *work);
1013 void kvm_arch_async_page_present(struct kvm_vcpu *vcpu,
1014                                  struct kvm_async_pf *work);
1015 void kvm_arch_async_page_ready(struct kvm_vcpu *vcpu,
1016                                struct kvm_async_pf *work);
1017 bool kvm_arch_can_inject_async_page_present(struct kvm_vcpu *vcpu);
1018 extern bool kvm_find_async_pf_gfn(struct kvm_vcpu *vcpu, gfn_t gfn);
1019
1020 void kvm_complete_insn_gp(struct kvm_vcpu *vcpu, int err);
1021
1022 int kvm_is_in_guest(void);
1023
1024 void kvm_pmu_init(struct kvm_vcpu *vcpu);
1025 void kvm_pmu_destroy(struct kvm_vcpu *vcpu);
1026 void kvm_pmu_reset(struct kvm_vcpu *vcpu);
1027 void kvm_pmu_cpuid_update(struct kvm_vcpu *vcpu);
1028 bool kvm_pmu_msr(struct kvm_vcpu *vcpu, u32 msr);
1029 int kvm_pmu_get_msr(struct kvm_vcpu *vcpu, u32 msr, u64 *data);
1030 int kvm_pmu_set_msr(struct kvm_vcpu *vcpu, u32 msr, u64 data);
1031 int kvm_pmu_read_pmc(struct kvm_vcpu *vcpu, unsigned pmc, u64 *data);
1032 void kvm_handle_pmu_event(struct kvm_vcpu *vcpu);
1033 void kvm_deliver_pmi(struct kvm_vcpu *vcpu);
1034
1035 #endif /* _ASM_X86_KVM_HOST_H */